Archive ouverte de Centrale Lyon
Not a member yet
32420 research outputs found
Sort by
Intelligence(s) : un webinaire pour explorer la thématique de la Fête de la science 2025: L'intelligence des arbres : mythe ou réalité ?
No full textNational audienceÀ l’occasion de la Fête de la science 2025, les coordinations régionale et départementales de la Fête de la science en Auvergne-Rhône-Alpes vous ont invité à un webinaire inédit consacré à la thématique « Intelligence(s) ». Deux heures d’échanges avec des chercheurs et médiateurs scientifiques, pour découvrir des aspects inédits de l’intelligence et nourrir vos projets de médiation.https://www.youtube.com/watch?v=i2fdXMQULcQAu programme, trois interventions et une session de questions réponses avec : Catherine LENNE, enseignante-chercheuse en botanique à l'Université Clermont Auvergne (UCA) et membre du laboratoire Physique et Physiologie intégratives de l'Arbre en environnement Fluctuant (PIAF). Catherine Lenne est ambassadrice de la Fête de la science2025 pour la région Auvergne-Rhône-Alpes. Nina GASKING, chargée de médiation scientifique de la Maison des Mathématiques et de l’Informatique (MMI) et Olivier DRUET, directeur de recherche CNRS, rattaché à l'Institut Camille Jordan, et directeur de la MMI. Caroline DEPECKER, rédactrice en cheffe du Cortex Mag et journaliste pour le prochain numéro du Pop’Sciences Mag « Intelligence »
Une recherche conception participative pour et avec un public d’étudiant·es avec TND (trouble neurodéveloppemental) : comment favoriser une participation « encapacitante » ?
No full textInternational audienc
: Conférence grand public "MathEnVille 2025"
No full textEn 1954, un mathématicien hors catégorie, tout à la fois athlète, prix Nobel d'économie, inventeur de jeux et surtout cerveau d'exception fait une découverte fracassante. Il met en lumière l'existence de surfaces à la fois grumeleuses et lisses ; une propriété inconcevable pour l'esprit humain. Aujourd'hui, des constructions explicites de telles surfaces permettent de résoudre le paradoxe de leur mystérieuse géométrie et révèlent une nouvelle classe d'objets : les fractales lisses
Disentangling temperature and Reynolds number effects in quantum turbulence
No full text(IF 9.1;Q1)International audienceThe interplay between viscous and frictional dissipation is key to understanding quantum turbulence dynamics in superfluid 4 He. Based on a coarse-grained two-fluid description, an original scale-by-scale energy budget that identifies each scale’s contribution to energy dissipation is derived. Using the Hall-Vinen-Bekharevich-Khalatnikov (HVBK) model to further characterize mutual friction, direct numerical simulations at temperatures 1.44 K ≲ T ≲ 2.16 K indicate that mutual friction promotes intense momentum exchanges between the two fluids to maintain a joint energy cascade despite their viscosity mismatch. However, the resulting overall frictional dissipation remains small (compared to the viscous dissipation) and confined to far-dissipative scales. This remarkable feature allows us to define an effective Reynolds number for the turbulence intensity in a two-fluid system, helping to disentangle the effects of Reynolds number and temperature in quantum turbulence. Thereby, simple physical arguments predict that the distance ℓ between quantized vortices (normalized by the turbulence integral scale L 0 ) should behave as ℓ / L 0 ≈ 0.5 Re κ − 3 / 4 with the Reynolds number based on the quantum of circulation κ . This law is well supported by a large set of experimental and numerical data within the temperature range of the HVBK model. Finally, this approach offers the possibility of revisiting the ongoing controversy on intermittency in quantum turbulence. It is shown that observed changes in intermittency arise from Reynolds number effects rather than from temperature variations, as proposed in recent studies
A General Sampling Framework for Graph Convolutional Network Training
No full textInternational audienceGraph Convolutional Networks (GCNs) have recently gained significant attention due to the success of Convolutional Neural Networks in image and language processing, as well as the prevalence of data that can be represented as graphs. However, GCNs are limited by the size of the graphs they can handle and by the oversmoothing problem, which can be caused by the depth or the large receptive field of these networks. Existing approaches address these limitations by leveraging minibatch training paradigm. However, the strategy of selecting subgraphs to form minibatches is a challenging task because of the dependency between nodes.In this work, we propose a general framework for generating minibatches in an effective way that ensures minimal loss of node interdependence information, preserves the original graph properties, and diversifies the samples for the GCN to improve generalization. We test our training process on real-world datasets with several well-known GCN models and demonstrate the improved results compared to existing methods
A gradient plasticity model for porous metals with random spheroidal voids: Theory and applications
No full textInternational audienceThis work deals with the development of a rate-independent, implicit gradient plasticity model for porous metallic materials comprising microstructures with an isotropic distribution of randomly oriented spheroidal voids. We take into account void shape effects via a single constant, the void aspect ratio, which can be used as a calibration parameter for the model. The non-local formulation introduces a characteristic material length, which serves as a regularization parameter and can be estimated by association to a microstructural dimension of the material at hand. The mathematical character of the resulting non-local problem and the conditions for loss of ellipticity are carefully examined. We show, both analytically and numerically, that the proposed model retains the elliptic properties of the governing equations and can provide mesh-independent numerical solutions in the post-bifurcation (softening) regime. This analysis also indicates that the critical localization strain is an increasing function of the void shape. Implementation of the model in the finite element software ABAQUS allows to investigate the effects of the various parameters through the numerical simulation of industrially relevant problems such as the cup-and-cone fracture of cylindrical bars and the Charpy V-notch test. By revisiting the first Sandia Fracture Challenge, we showcase the capability of the model to sufficiently reproduce real-world experimental results while maintaining a manageable number of calibrated parameters
Développement et pilotage d’un test de conscience phonologique utilisable en autonomie
No full textInternational audienceLa conscience phonologique est définie comme « one’s ability to recognize, discriminate, and manipulate the sounds in one’s language, regardless of the size of the word unit that is the focus » (Anthony & Francis, 2005). La conscience phonologique, et plus particulièrement la conscience phonémique, joue un rôle crucial dans l'acquisition de la lecture en L1 et L2 (Snow et al., 2001 ; Haigh et al., 2011), de la prononciation en L2 (Kivistö-de Souza, 2015), ainsi qu'en compréhension de l'oral en L2 (Li et al, 2012). En outre, plusieurs auteurs défendent l'hypothèse selon laquelle la conscience phonologique n'est pas liée à une langue spécifique, mais plutôt à une méta-compétence linguistique générale (Comeau et al., 1999 ; Cummins, 1979). Dans la littérature scientifique, les études sur la conscience phonologique utilisent soit des tests ad hoc difficiles à utiliser dans d'autres contextes, soit des tests commerciaux déjà validés et calibrés auprès d'une population native (par exemple CTOPP, Wagner, Torgesen & Rashotte, 1999). L'utilisation de ces tests avec des enfants apprenants L2 s'avère difficile, en raison d'un manque de connaissances lexicales. De plus, ces tests sont généralement sur papier et conçus pour être administrés individuellement par des professionnels. Ils sont donc coûteux, longs à utiliser et difficiles d'accès pour les enseignants et les chercheurs.Le test développé dans le cadre des projets Trans3 et C-maphore comprend 4 tâches (segmentation en phonèmes, repérage de syllabe initiale et de rime, fusion de syllabes). Chaque tâche est composée d'items en anglais, français et taïwanais. Pour que le test soit utilisable en autonomie par des enfants, nous avons fait différents choix de développement (contextualisation, éléments graphiques motivants, pas de réponse orale attendue, etc.)Nous présenterons les premiers résultats issus d'une étude menée auprès de 114 élèves de GS et CP : corrélation avec un test reconnu (validation de critère externe), fiabilité du test, corrélation entre les différentes tâches et entre les items de différentes langues.Notre étude permet de contribuer à la compréhension des caractéristiques de la conscience phonologique et ambitionne de mettre un nouvel outil à la disposition des enseignants et des chercheurs
Minimax testing in a statistical inverse problem with unknown operator
No full textWe study minimax testing in a statistical inverse problem when the associated operator is unknown. In particular, we consider observations from an inverse Gaussian regression model where the associated operator is unknown but contained in a given dictionary B of finite cardinality. Using the non-asymptotic framework for minimax testing (that is, for any fixed value of the noise level), we provide optimal separation conditions for the goodness-of-fit testing problem. We restrict our attention to the specific case where the dictionary contains only two members. As we will demonstrate, even this simple case is quite intrigued and reveals an interesting phase transition phenomenon. The general case is even more involved, requires different strategies, and it is only briefly discussed
Verifying output amplitude constraints in a robust manner for an identification experiment with a multisine excitation
No full textInternational audienceWhen designing an identification experiment to determine a model of a given true system, a frequent requirement is to ensure that the amplitude of the output of the true system will not exceed a certain value. An important issue when dealing with such an output amplitude constraint is that the output depends on the unknown true system. In this paper, we circumvent this issue by considering an uncertainty region containing the true system. This uncertainty region is assumed to have been delivered by an initial identification experiment. We then provide an LMI procedure allowing to verify that, when subject to a given multisine excitation, the output of all systems in the uncertainty region satisfies the output amplitude constraint. We also provide an LMI procedure to address another frequent requirement in system identification i.e., that the amplitude of the multisine excitation does not exceed a certain value
Tribological behavior and microstructure of IN718 produced by SPS in comparison with LMD and forged grades
No full textInternational audienceSpark Plasma Sintering (SPS) was evaluated for its potential to enhance the wear resistance of Inconel 718, a key material for copper extrusion dies. Comparative wear tests assessed the tribological and mechanical performance of SPS-produced parts versus Laser Metal Deposition (LMD) and wrought parts. The study reveals that SPS, especially with ball-milled powders followed by aging, yields materials with significantly higher hardness and comparable wear resistance to LMD, despite distinct microstructures. Forged materials demonstrated superior performance, attributed to their nearly defect-free nature. The study highlights the influence of carbide stability and distribution in SPS materials as a promising route for optimization. These findings underscore SPS's ability to compete with traditional methods, offering tailored properties for high-demand applications and potential cost-efficiency improvements.</div